07.4.21.23:23: SINO-ARABIC PARALLELS 3: MÁS-RII
That's más as in Spanish 'more'. David Boxenhorn has been finding more Semitic data that seem to strengthen the typological foundation of the Emphatic Sinitic Project (ESP!). I haven't had time to examine these three PDFs carefully yet because I'm still reeling from Islam Youssef's two-pager.
Five years ago, when I knew very little about Arabic and was totally ignorant of Cairene Arabic and other nonstandard dialects, I gave a talk at the University of Oregon in which I hypothesized that Norman's (1994) 'emphasis' in Old Chinese was partly caused by lost word-initial syllables that left phonetic traces in the surviving syllables.
Two years ago, when I was still in the dark about colloquial Arabic, I refined my 2002 ideas into an article that is still awaiting publication. In it, I proposed that Old Chinese had a process that I called 'syllabic harmony'. Monomorphemic words were 'harmonized' in terms of 'emphasis': they were either 'emphatic' or 'nonemphatic' from beginning to end:
|Type of word||'Emphatic'||'Nonemphatic'|
|Sesquisyllable (monosyllable plus unstressed presyllable)||*Cʌ-CV||*Cɯ-CV|
(OC syllables could be quite complex in structure, and *CCʌ-/*CCɯ- presyllables may have existed. I am using 'CV' only for demonstration purposes.)
Both my 2002 and 2005 'emphatic harmony' proposals were inspired by vowel harmony in the neighboring 'Altaic' languages (more on them in the future). The phonetic values of the 'emphatic' and 'nonemphatic' presyllables were directly inspired by the Middle Korean vowels ʌ (arE a, the minimal 'yang' vowel) and ɯ (으, the minimal 'yin' vowel).
I had no idea that there was anything like this in Arabic until I saw Youssef's examples of 'emphasis spread' (ES) on the second page of his "Dorsal Harmony in Cairene Arabic". At last, a phenomenon like 'emphatic harmony'* in a living language!
Now I see why a visiting Arabic linguist liked my 2002 talk. She must have recognized that I was talking about ES. After my talk, she asked me for my contact information and I gave it to her without hesitation. Unfortunately, I didn't ask for hers, and I've never heard from her since. A shame, but I've enjoyed trying to work out this problem with David's assistance. The best is yet to come ... I hope.
Next: "The Butterfly Case", part tr-ES.
*Note that Cairene Arabic ES does not affect all words from beginning to end:
[Emphasis s]preading from left to right is interrupted by a subset of vowels and consonants: the targets are all segments except non-tautosyllabic [I, ii, ee, zh], which constitute the set of blockers.
Old Chinese may also have had its own 'blockers'. This might explain why there are exceptions to 'emphatic harmony'. However, at this point, I have no idea what the blockers might have been, though I could test some guesses (Late Old Chinese *zh and *z?**) based on the precedent of Cairene Arabic.
I realize that it's odd to refer to a modern language as a 'precedent' for the reconstruction of a dead language.
**Both of these consonants occurred exclusively before 'nonemphatic' vowels, so it's possible that they prevented their syllables from 'harmonizing' with any preceding (pre)syllable:
Sesquisyllable: *Cʌ-z(h)V (did not become *Cʌ-z(h)V)
Disyllable: *CV.z(h)V (did not become *CV.z(h)V)
Although *zh- did not exist in earlier Old Chinese, *z- might have existed as an allophone of /s/ before *l- and *y-. Earlier Old Chinese*zl- and *zy- were reduced to *z(y)- in Late Old Chinese.
07.4.20.23:59: SINO-ARABIC PARALLELS 2: MASRII
I was disappointed when I saw the Wikipedia article on Egyptian Arabic (masrii*) because it said nothing about vowel allophony. I was hoping to find more parallels with Old Chinese. The transcriptions do, however, imply that
a > [ɑ] after 'emphatics'
including 'emphatics' that are not even adjacent to the vowel: e.g.,
sɑhbɑ 'friend' (f.) < presumably something like Classical Arabic /sahiba/
In the above case, 'emphasis' seems to have 'jumped across' the non-'emphatic' b. Or has it?
If my understanding of Islam Youssef's analysis is correct, that entire word might be 'emphatic': sɑhbɑ
One interesting characteristic about Cairene Arabic inventory is that all segments [even /y/!] except [q] have emphatic counterparts, but some of these are found only in very restricted environments. All consonants are necessarily emphatic in a syllable containing an emphatic vowel [ɑ]. However, of all the emphatic sounds, only some coronal consonants [d, t, s, z, r] can appear in environments other than [ɑ] (non-low vowels).
So far, Cairene Arabic comes closest to my Old Chinese reconstruction in which all consonants but */y/ had 'emphatic' counterparts**. However, in my Old Chinese reconstruction, the 'emphatic'/non-'emphatic' contrast was found before all six vowels */a i u e o ə/ and not just before the allophones of /a/, *[æ] and *[ɑ] (*[ɑ]?): e.g. (examples added 4.21.10:26),
睹 *ta' [tɑ] 'see' : 者 *ta' [tæ'] 'person'
顛 *tin [tIn] 'top of head' : 眞 *tin [tin] 'true'
老 *ru' [rU'] 'old' : 柳 *ru [ru] 'willow'
雞 *ke [qE] 'chicken' : 支 *ke [ke] 'branch'
鉤 *ko [qO] 'hook' : 俱 *ko [ko] 'together'
改 *klə' [qlʌ'] 'change' : 己 *klə' [klə'] 'self'
Next: ES in CA and OC.
*07.4.21.00:09: Why does the a (emphatic [ɑ]?) of masrii 'Egyptian' correspond to i of standard Arabic misrii?Is this a case of what Youssef would call 'leftward spread' (more about this next time)? Is this what happened over time?
Stage 1: misrii (emphasis only in s)
Stage 2: misrii (emphasis spread to preceding i)
Stage 3: masrii (short emphatic vowels merge with a)
Stage 4: masrii (emphasis spread to preceding m)
07.4.21.8:49: The Wikipedia IPA transcription of مصري is [mɑsˁɾɨ]. This implies that /i/ backs to a central vowel [ɨ] after /r/ (Youssef's /r/? - and after 'emphatics'?). I don't think this happened in Old Chinese. I assume that OC */i/ lowered to *[I] after 'emphatics'.
**07.4.21.8:54: See my list of Old Chinese consonants here. If there was an emphatic *y (as in Cairene Arabic) it may have merged with emphatic *l in Late Old Chinese.
07.4.19.23:59: SINO-ARABIC PARALLELS 1: DARIJA
I've been looking at Wikipedia articles on Maghrebi Arabic (Darija 'dialect') in search of parallels with Old Chinese historical phonology.
This morning, I presented what I considered to be the first stage of vowel lowering after 'emphatic' onsets:
In the second stage, these vowels lowered even further:
OC */i/ > *[I]
OC */u/ > *[U]
OC *[I] > Late OC *[EI] (Schuessler's *ei; also cf. Pulleyblank's Early Middle Chinese *Ey)
OC *[U] > Late OC *[OU] (Schuessler's *ou)
This is not unlike what happened to *ii and *uu in Moroccan and Algerian Arabic after 'emphatics':
*ii > e
*uu > o
In Libyan Arabic, /i/ and /u/ lower to [e] and [o] in the present and imperative forms of
Old Chinese /a/ after non-'emphatics' ended up as a mid vowel in Middle Chinese dialects (e.g., *ö) and as a high vowel in modern Chinese languages: e.g., standard Mandarin ü.
roots with initial uvular, pharyngeal and glottal [!] phonemes (namely X H h [!] R ` ' [!], but not [uvular] q) [or the dental 'emphatics'!]
Similarly, /aa/ after non-'emphatics' can raise to EE or even ee in Tunisian Arabic.
(Tunisian Arabic /aa/ backs to ɑɑ after 'emphatics'. Cf. my reconstruction of *[ɑ] for Old Chinese */a/ after 'emphatics'.)
Tangut also underwent an *a > e shift in some words. See Matisoff (2004), which also lists examples of *a becoming i and u in Tangut. Note, however, that raising in Tangut and other Qiangic languages had nothing to do with a lack of 'emphasis' (though postvelars did block raising).
The parallels between Darija and Old Chinese are not simply vocalic:
In Moroccan Arabic, non-'emphatic' /t/ "is pronounced with noticeable affrication, almost like /ts/'. This is reminiscent of the affrication of Old Chinese non-'emphatic' dental stops in Late Old Chinese:
OC *t, *th, *d > LOC *ch, *chh, *j
Tunisian Arabic has a "rare" 'emphatic' phoneme n like my Old Chinese *n (which would not have been rare*). This 'emphatic' nasal is also "marginally" phonemic in Hassaniya Arabic.
Next: Eastward to Egypt.
*07.4.20.00:50: Examples of *n-words from the index of Pulleyblank's Outline of Classical Chinese Grammar (1995: 179; the reconstructions are mine):
迺 / 乃 *nəng' 'you, your' (cf. 戎 *nung 'you' and other second person pronouns with non-'emphatic' *n)
乃 *nəng' 'then'
奈 *nats 'like' (Pulleyblank derived this from 若之 *nak tə, but the initials don't match and *-kt > *-ts seems strange [unless *t had already palatalized or become a palatal affricate: *-kty ~ *-kch > *-ts.])
難 *nar 'difficult', *nars 'difficulty'
能 *nəng 'be capable'
寧 *neng 'quiet; rather'
Other common *n-words were
南 *nəm 'south'
男 *nəm 'man'
腦 *nu' 'brain'
內 *nəps 'inside' (cf. 入 *nəp 'enter')
年 *nin (< **-ng) 'year'
念 *nems 'think'
農 *nung 'agriculture'
暖 *non' 'warm'
07.4.18.23:59: LESS THAN UBYKH-UITOUS?
Guillaume Jacques led me to these Ubykh sound files. Some have handwritten phonemic (not phonetic!) transcriptions: e.g.,
for [faaXye t'qw'o-kwobzhæ] (to my ears)
'once two-man' = 'Once two men ...'
Wikipedia has a partially typed version of the transcription of "Eating fish makes you clever" (.wav), a story told by the late Tevfik Esenç, the last speaker of Ubykh. The above example consists of the first two words of "Eating fish".
I was expecting to find a lot of pharyngealized consonants, but in fact the only one that appears in the typed transcription is b (in the word /shwəb/ 'bread'). Since this story is not a Ubykh version of La Disparition, I assume that the low frequency of pharyngealized consonants is normal. For comparison, look at the first poem of Shijing. It may have had 'emphatics' in seven of its sixteen syllables:
*[q`R`On q`R`On tshæ ku]
'(sound of ospreys) osprey'
*dzə' gay tə tu
*[dz`ʌ' G`ɑy tə tu]
'be-on river -'s islet'
*'iw'-liw' duk rna'
*[`Iw l`Iw duk rnæ']
'modest virtuous woman'
*kun-tsə' xu' gu
'prince good mate'
*[kun tsə' X`U' gu]
Is there any language in the world with such a high frequency of 'emphatics'?
07.4.19.8:13: In my phonetic reconstruction of the poem, vowels after 'emphatics' are lowered with the exception of /a/ which is backed since it is already low:
|+'emphatic' onset||*[I] (lowered)||*[U] (lowered)|
|+'emphatic' onset||*[E] (lowered)||*[ʌ] (lowered and backed)||*[O] (lowered)|
|-'emphatic' onset||*[æ] (fronted, or just [a]?)|
|+'emphatic' onset||*[ɑ] (backed)|
I am not sure whether /a/ was really fronted after non-'emphatics'.
07.4.17.23:59: UBYKH-UITOUS EMPHATICS?
Although I've been advocating the hypothesis that Old Chinese had 'emphatics' for seven years*, I still have my doubts. I would like to see a lot of typological parallels with languages which definitely do have 'emphatics' before I am totally sold on it.
In my reconstruction of Old Chinese, almost all consonants belong to 'emphatic'/non-emphatic pairs:
|*p||*p||*t||*t||*ts||*ts||*k||*k [q]||*kw||*kw [qw]||*'||*' [`]|
|*ph||*ph||*th||*th||*tsh||*tsh||*kh||*kh [qh]||*khw||*khw [qhw]|
|*b||*b||*d||*d||*dz||*dz||*g||*g [G]||*gw||*gw [Gw]|
|*m||*m||*n||*n||*ng||*ng [N]||*ngw||*ngw [Nw]|
|*hm||*hm||*hn||*hn||*hng||*hng [N]||*hngw||*hngw [hNw]|
|*s||*s||*x||*x [X]||*xw||*xw [Xw]|
|*w||*w||*l||*l||*r||*r (uvular [R]?)||*y||'Emphatic' velars were probably uvulars. [N] here is a uvular nasal and not a retroflex nasal [N].|
|*hl||*hl||*hr||*hr (uvular [XR]?)|
The only exception is *y which I've reconstructed under the influence of Schuessler (2007: 96-99). It has no 'emphatic' counterpart *y. (Underlining indicates 'emphasis'.)
Until recently, I did not know of any language that had so many 'emphatics'.
only had five 'emphatics': *θ, *t,
*s, *hl (lateral
fricative), and k (=
q). I suppose that `
h would count as
premodern Hebrew, the 'emphatics' were t
and s and the 'emphatic'-like
consonants were q, `, and
In modern standard Arabic, the 'emphatics' are all dental and the 'emphatic'-like consonants that have similar effects on vowels are uvular or pharyngeal with the exception of /r/ (Kaye in Comrie [1987: 670]). According to Kaye,
In Old Arabic, the primary emphatics were, in all likelihood, voiced.
There are no emphatic nasals and the emphatic l is restricted to the name of God ('allaah).
Kaye pointed out that "[s]ome modern Arabic dialects have many examples of /l/" and he also noted an emphatic p in Iraqi Arabic.
That was as far as my knowledge of 'emphatics' went - until Sunday night.
I got an email from David Boxenhorn, who found a reference to "secondary" (what does this mean?) Moroccan Arabic emphatics b, m, r, z, and g (< emphatic-like *q according to Wikipedia).
And I finally found out why Ubykh "has the largest consonant inventory of all documented languages which do not use clicks". I had known that Ubykh had a huge number of consonants, but I didn't know what those consonants were until now. If 'emphatic' is defined as 'pharyngealized consonant' (which is not the only possible definition), Ubykh has
- a nearly full set of 'emphatic' labials: p, b, p', m, w, v
(but f has no 'emphatic' counterpart, and plain v is only in loanwords!)
- 'emphatic' ejectives: p', q', q'w-
- 'emphatic' uvulars distinct from nonemphatic uvulars: q, q', q'w, X, Xw, R, Rw-
- 'emphatic' labialized uvulars: Xw, Rw
- but no 'emphatic' y (which Old Chinese also lacked)
Now many Old Chinese 'emphatics' have counterparts somewhere in some language known to me. The exceptions are:
- 'Emphatic' aspirates: *ph ([pX]?) etc.
- A distinction between 'emphatic' fricatives and affricates: *s vs. *ts. (In Hebrew, earlier s became [ts].)
- Voiceless sonorants: *hm ([Xm]?) etc.
- Dental and velar (uvular?) nasal 'emphatics': *n, *ng, etc.
(I don't know of any language with initial uvular nasals.)
Such consonants are not a priori impossible, but I'd still like to verify that they exist in actual languages.
The biggest surprise of the Wikipedia Ubykh phonology article (not the best source, I know) is the absence of any mention of vowel allophony after pharyngealized consonants. As one would expect, labialized consonants trigger vowel backing and rounding, and palatalized consonants trigger vowel fronting. But do pharyngealized consonants really trigger nothing, or was their effect simply omitted? If Ubykh /a/ and /ə/ have no allophones after pharyngealized consonants, then perhaps Old Chinese vowels originally had no allophones after pharyngealized consonants: e.g., the Shijing rhyme which I interpreted as
蒲 *ba [bˁɑ] 'cattail' : 居 *ka [kæ] 'dwell'
in "Something's Fishy" would have been a perfect rhyme
蒲 *ba [bˁɑ] 'cattail' : 居 *ka [ka] 'dwell'
and the variation of /a/ developed later.
*07.4.18.7:59: Leon Serafim first brought Norman's (1994) hypothesis to my attention around 1995. But I didn't take it seriously until I looked at a description of Maltese phonology five years later. Maltese vowel allophony was reminscent of the behavior of Chinese vowels after Norman's proposed pharyngealized Old Chinese consonants. I then looked at Arabic vowel allophony and became increasingly convinced, though to this day I am still not entirely sure.
07.4.16.23:59: THE FIRST PARASINOGRAPHY?
Is the "Hun-Syanbi script" for real?
The Chinese hypothesis connects the [Orkhon Turkic] script to the reports of a 2nd century BC renegade Chinese dignitary named Yue [what's the graph?] who
"taught the Chanyu (rulers of the Hsiung-nu [= Xiongnu]) to write official letters to the Chinese court on a wooden tablet 31 cm long, and to use a seal and large-sized folder." [What's the source of this quote?]
The same sources tell that when the Xiongnu noted down something or transmitted a message, they made cuts on a piece of wood ('k'o-mu'), and they also mention a "Hu [i.e., 胡 Md hu 'barbarian'] script". At Noin-Ula and other Hun burial sites in Mongolia and region north of Lake Baikal among the artifacts were discovered over twenty carved characters. Most of these characters are either identical or very similar to the letters of the Turkic Orkhon script.
Here are two samples of the 'Hun-Syanbi' script. I didn't even know it existed until tonight. Neither EG Pulleyblank nor Alexander Vovin mentioned it in their work on the Xiongnu language.
The name 'Hun-Syanbi' implies that the script was used by the Xianbei (= 'Syanbi') as well as the Xiongnu (= 'Huns'), though the Wikipedia article gives no evidence for this.
Moreover, I am not sure the 'Hun-Syanbi' script is actually a script. Are there actual texts in this script? Could these simply be random attempts to imitate the shapes of sinographs? Could the resemblance to the later Orkhon Turkic runes be totally coincidental? After all, some of the 'Hun-Syanbi' characters happen to resemble Roman letters (e.g., X, V, T, S, I, O, Y, P, M). The shared use of simple shapes is meaningless unless those shapes also have similar phonetic values. Does anyone know what sounds the 'Hun-Syanbi' characters represented?* Are the Chinese characters next to the 'Hun-Syanbi' characters modern attempts to guess their meanings and/or sources? Could the 'Hun-Syanbi' characters be meaningful yet nonlinguistic tamghas?
For now, I'll continue to consider the Khitan Big Script (megakhitanography?) to be the first parasinography** (a Chinese-like or sinoform script, as opposed to a local adaptation of sinography).
Coming soon: Why do fish swim among the runes?
*07.4.17.1:45: I doubt it, because the decipherment of a Xiongnu script would resolve the old controversy of the affiliation of the Xiongnu language. Both Pulleyblank and Vovin believe that Xiongnu was a Yeniseian language.
Although Xiongnu contained Turkic-like titles, these titles were presumably borrowed into Turkic from Xiongnu and are not necessarily evidence of a genetic relationship between Xiongnu and Turkic. Cf. the spread of Latin Caesar into non-Italic languages: e.g., Dutch keizer, German Kaiser, and Russian tsar'. (More cognates here.) Of course, all four of these languages are related, whereas Xiongnu and Turkic were not related.
**07.4.17.1:55: Janhunen (1994) proposed that the Khitan Big Script was not a 10th century Khitan invention but an adaptation of an preexisting variant of sinography. However, no such variant has been documented to the best of my knowledge.
07.4.15.23:59: SOMETHING'S 魚 FISHY (PART 1)
In last week's posts, I kept changing my mind about how to reconstruct the premodern Chinese 魚 'fish' rhyme category1. Many of the sinographs used to transcribe the languages of the Korean peninsula and the Japanese archipelago represent syllables of this rhyme category. (Both syllables of the transcription 夫餘 for ?*bala [see here] belong to the 'fish' category.) Therefore the proper reconstruction of the 'fish' rhyme is crucial not only for the history of Chinese but also for the history of its linguistic neighbors. Although I may change my mind again about the reconstruction of 'fish', I am writing this series of posts to document my current thoughts about it.
The earliest value of 'fish' was *-a2. Three lines of evidence point toward this value for Old Chinese (OC):
1. Internal: Alternations with other *-aC rhymes (Schuessler 2007: 18):
於 OC *'a 'in' : 焉 OC *'an 'in it'
無 OC *ma 'not have' : 莫 OC *mak 'none'
無 OC *ma 'not have' : 亡 OC *mang 'lose'
(The 'fish' rhyme words are roots and the -aC forms are their derivatives.)
2a. External: Cognates (genetic or otherwise)
魚 OC *nga 'fish' : Written Tibetan ña < *ngya 'fish'
(ngy- is not possible in Written Tibetan)
(See Schuessler 2007: 586-587.)
無 OC *ma 'not have' : Written Tibetan ma 'not'
(See Schuessler 2007: 518-519.)
苦 OC *kha' 'bitter' : Written Tibetan kha, Written Burmese khaH 'id.', Tangut
TT1791 BITTER kha 2.14
(gloss from Gong 1995: 423)
五 OC *nga' 'five' : (borrowed into?) Written Tibetan lnga, Written Burmese ngaH 'id.'; borrowed into Tai: e.g., Siamese ห้า haa < *hng-? 'id.'
TT3119 ŋwə 1.27 'five' with -wə instead of -a may be due to vowel alternation, though the reason for the -w- is unknown.)
弩 OC *na' 'crossbow', borrowed from some Austroasiatic form like Proto-Viet-Muong *s-naa' 'crossbow' or Khmer snaa 'crossbow'
(See Schuessler 2007: 404-405.)
2b. External: Transcriptions:
烏弋山離 OC *'a lək ksan ray 'Alexandria'4 (Han shu 96)
(See Pulleyblank 1962: 116, 128.)
浮 屠 OC *bu da 'Buddha'
(the graphs literally mean 'floating butcher'!)
I think that this *-a had two allophones:
a back allophone [ɑ] after 'emphatic'(-like) consonants (written on this blog with underlining):
e.g., 古 OC ka' [qɑ'] 'old'
a front allophone [æ] after all other consonants:
e.g., 居 OC ka(') [kæ('/s)] 'dwell'
(for the final glottal stop and *-s, see Schuessler 2007: 322; Schuessler reconstructed *-h instead of *-s)
Cf. the allophony of standard Arabic /a/ after 'emphatic'(-like) and non-'emphatic'(-like) consonants (Kaye in Comrie [1987: 670]):
قَ /qa/ = [qa]
كَ /ka/ = [kæ]
(But note that unlike Arabic script, sinography uses the same phonetic elements for syllables with both 'emphatic'[-like] and non-'emphatic'[-like] initials: e.g.,古 in both 古 *[qɑ'] and 居 *[kæ(')] . This may either be an argument against Norman's  emphatic hypothesis, or an indication that 'emphasis' had not yet become phonemic when sinography was initially invented.)
This allophony was not initially a barrier to rhyming. In Shijing, the two types of -a (*[ɑ] and *[æ]) interrhymed freely (Starostin 1989: 561-564): e.g.,
蒲 *ba [bˁɑ] 'cattail' : 居 *ka [kæ] 'dwell'
(It is also possible that the Shijing dialect[s] lacked this allophony.)
However, in later poetry, these 'fish' would go their separate ways.
1The word 魚 'fish' has always belonged to this rhyme category. Premodern Chinese rhyme categories are always named after one of their members.
2For the purposes of this post, I will treat *-a' and *-as (> later *-ah) as part of the 'fish' rhyme.
Some reconstruct a velar or glottal consonant after *-a: e.g., *-ag (Li 1971) or *-aH (Pulleyblank 1962). However, there are no cognates indicating a final back consonant, and sinographs for 'fish' syllables do not represent foreign syllables ending in *-g, etc. Thus I do not reconstruct any final consonants in the 'fish' category. See Gong (1995: 57-59) for an argument in favor of Li's *-g.
307.4.16.1:04: I am not certain that Gong's gloss is correct. His BITTER represents the first half of a disyllabic word
TT1791 1792 kha 2.14 rerw 1.87 (Pearl 151)
glossed by Grinstead (1972) as 'a vegetable' and in Nishida (1964: 201) as 'bitter - (ginko [sic] nut)'. If TT1791 were an adjective, it should appear in second, not first position: (name of plant) + BITTER, not BITTER + (name of plant). It is, however, possible that the word is a calque of Chinese 苦蕖 Ixeris chinensis (lit. 'bitter ?'). (蕖 is usually the second half of the disyllabic word 芙蕖 'lotus'. I don't know what it means in 苦蕖, and apparently neither did Nishida.)
I cannot find any other Tangut word meaning BITTER with a velar initial. The words for BITTER listed by Grinstead are
TT0407 zar 1.80
translated by Nevsky (1960 II: 652) as 辛 'bitter' and горький 'bitter' and
TT3190 syịy 2.54 BITTER/CYCLICAL SIGN
which may be a loanword from Chn 辛 'bitter', possibly pronounced sĩ in the northwestern dialect known to the Tangut. Nevsky (1960 II: 469) translated it as '辛 acrid'.
As far as I know, TT3190 is not used as a cyclical sign. The eighth Heavenly Stem 辛 (written with the same sinograph as 'bitter') corresponds to Tangut
TT2642 khieey 1.38
which shares a kh- with the other Chinese word for 'bitter' (苦 OC *kha') but nothing else. The similarity in initials may be coincidental. Tangut -ieey is not known to come from an earlier *-a.
4Perhaps the great vowel warping which I will discuss in depth next time had already begun at this point. If so, then 烏弋山離 was OC *'a lïək ksan rïay.
離 is phonetic in 魑 Middle Chinese *Thiə < OC ?*thray5 'demon' with a retroflex stop initial. Perhaps 離 represented OC *təray with a dental stop initial which would be even closer to Greek -δρεια.
烏 弋山離 may represent either Alexandria Prophthasia or Alexandria Arachoton (modern Kandahar in Afghanistan). Details here. See both on this map.
There are many other Alexandrias which were not designated by 烏弋山離. The Alexandria of Egypt is Rakotə in Coptic. I assume that Rakotə is unrelated to Ἀλεξάνδρεια.
5Sagart (1999: 28) would derive retroflex *Th- from Old Chinese *hr-. This fortition of *hr- to a stop *Th- has no parallels among the other nonemphatic OC sonorants which do not become later stops in his reconstruction:
Hence I prefer to derive *Th- from an OC cluster *thr-.
Examples of nonemphatic OC *hr- > MC *x- from Schuessler (2007: 57):
脅 OC *hrap > MC *xïap
畜 OC *hruk > MC *xuk
(Emphatic OC*hr- also became MC *x-. Note the phonetic similarity between a uvular r and the uvular voiceless fricative [X].)
But according to Schuessler (2007: 123), OC *hr- can also become MC *Th- (as in Sagart's reconstruction). When did OC *hr- become a fricative, and when did it become a stop?
6Should *sh- < *hng- be reinterpreted ae *sh- < *s-hng-? E.g.,
勢 OC *s-hngets > Late OC *shies > Middle Chinese *shieyh 'force'
Sagart (1999: 69) did not propose what OC *s-hng- would become in Middle Chinese.
07.4.16.1:39: Schuessler (2007) reconstructed *hng- as the source of later *x- and *ngh- as the source of later *sh-. (See pp. 123 and 570-571 for examples.)
I don't know what the difference between *hng- and *ngh- is. Is *hng- voiceless and is *ngh- a voiced aspirate? Schuessler does not reconstruct such a distinction for other OC sonorants: e.g., *hy- : *yh-. Does any extant language have such a distinction?